license: cc-by-nc-4.0
tags:
- climate
- trees
- images
size_categories:
- 1M<n<10M
The Curio Tree Dataset
This dataset contains much of the tree inventory, images and stories data that was collected on the Curio platform before it was sunset. The data was extraced from a number of database tables and includes;
- The inventory details of 2.5 millions trees from locations across the globe (location, species, diameter at breast height (DBH), height, vitality etc, where available)
- 27,288 images of trees that were uploaded onto the platform by our community and linked to individual trees and their species information etc.
- Notes (stories), tags and conversations linked to trees.
Dataset Description
Curio was an environmental education and outreach platform that was predominantly focused on urban forestry. It connected the various stakeholders involved in the management of urban forestry with the public and importantly made all data uploaded via its web and mobile apps publicly available. The platform was live from March 2016 until August 2023 when the maintainence overheads made its ongoing availability infeasible. Curio was supported in its early stages by two European Space Agency projects, through the New Commons and Curio Canopy. A sense of the platform and how it worked can be found via the videos on its supporting youtube channel
This repository contains much of the tree inventory, images and stories data that was collected on the platform via our community, projects we helped support and open data tree inventories we uploaded onto the platform. We are keen to make this data available for research purposes in the hope it might be of benefit to others and to further the efforts of our community.
We have endeavored to name as many of those great projects and data sources that were hosted on the Curio platform in the attribution section below. If there are any omissions or errors please contact us.
A related project involved generating a high resolution map of tree canopy cover for the Greater London Authority. Details of that project and dataset can be found on the London Datastore Curio Canopy page.
- Curated by: Breadboard Labs
- License: cc-by-nc-4.0
Dataset Sources and Attribution
Many people picked up the app and contributed to the data that was collected. Curio was also used to support many great projects and initiatives. We have endeavoured to mention many of those projects below along with the open data tree inventories we uploaded onto the platform.
Collaborative projects supported by Curio
Dublin City Council’s Parks, Biodiversity and Landscape Services & School of Geography at University College Dublin - Tree Mapping Dublin
Municipality of Oslo Agency for Urban Environment - Inventory and ecosystem services report hosting
Mountshannon Arboretum - Forester Bernard Carey initiated the Mountshannon i-Tree project, in conjunction with UCD and UK-based consultancy Treeconomics.
SLU - Alnarp - Skåne Tree Inventory and support for and involvement in the New Commons and Curio Canopy projects
Malmö Stad - Malmö Tree Inventory and support for and involvement in the New Commons and Curio Canopy projects
Open Data Sources Attribution
The Greater London Authority Datastore - Local Authority Maintained Trees
NYC OpenData - 2015 Street Tree Census - Tree Data
Open Data BDN - Street trees of the city of Barcelona
Open Data Bristol - Trees
Open Data NI - Belfast City Trees
Denver Open data - Tree Inventory
Open Data DK - City of Copenhagen Trees
Palo Alto Open Data - Palo Alto Trees
Fingal County Council Open Data - Fingal County Council Trees
Data SA - City of Adelaide Street Trees
Open Data Boulder Colorado - Tree Inventory Open Data
Biodiversity Ireland - Hertitage Trees Ireland
Birmingham City Council Trees
Uses
The data is free to be used for research purposes subject to the cc-by-nc-4.0 licence and suitable attribution, please see the citation section below
Some potential uses might include;
- Investigations into urban tree biodiversity.
- The development of algorithms for extracting tree attributes via photos or streetview imagery.
- A tree species detection app.
- The detection trees of via satellite imagery.
- Species identfiication via hyperspectral tree.
It worth noting that for most use-cases cleaning, analysis and processing of data will be necessary. The completeness of tree inventory data varies greatly and users were not directed in anyway in terms of how to frame the photos they took and uploaded via the Curio app.
Dataset Structure
TaggedTrees
Number of data points: 2,593,139
The details of an individual tree including its location, species, diameter at breast height (dbh), vitality etc. when available
Images
Number of data points: 27,288
The details of images that were uploaded to the platform. The path to the actual image uploaded, this can be found in uploads directory. The details of what the image was attached to which usually was a ‘Story” that was then attached to a tree are also included.
Uploads:
The set of images referenced in the images data file. The set of images was quite large even when zipped and so was broken up into 10gb chunks. Download each of the chunks and then run unzip on the uploads.zip file
A folder containing downsized versions of the images based on a fixed width has also been included - resized-uploads-width1200.zip
Stories:
The details of a story that was attached to tree
Notes:
The text included in a story/note about a tree.
Conversations & Comments:
Comments grouped by conversations linked to a particular Story
TreeSpecies
The tree species dictionary we built to support the platform. Each TaggedTree has a tree_species_id that references an entry in this dictionary when populated.
TreeSpeciesAliases
The local names across multiple languages that can used to describe a species of tree contained in the TreeSpecies dictionary
Tags and Taggings
Trees could be tagged with details such as diseased, monitored, newly_planted, apples, overhead cables etc. Anything at all really that could later be used to filter, group or identify trees of interest as well describe their state.
Dataset Creation
Curation Rationale
The goal of the Curio platform was to educate, engage and democratised access to environmenatal information. Making the data collected on the platform available in this form is seen as an extension of that mission.
Data Collection and Processing
All data was collected via the Curio app by its community. Where inventory data was uploaded in bulk we preprocessed the data to ensure details such as species information where mapped to the species dictionary we deinfed and that has been included in this release.
Before making the data available on this platform we decided to run face detection and blur any obvious, detectable faces found in the images that have been included.
Citation [optional]
@misc{CurioTreeData, title = {The Curio Tree Dataset}, author = {Conor Nugent and Paul Hickey}, year = {2023}, publisher = {HuggingFace}, journal = {HuggingFace repository}, howpublished = {\url{https://https://huggingface.co/datasets/BreadboardLabs/CurioTreeData}}, }
Dataset Card Authors
Conor Nugent and Paul Hickey